Evolutionary resurrection of flagellar
motility via rewiring of the nitrogen
regulation system
Article
Accepted Version
Last edited version for re-submission
Taylor, T. B., Mulley, G., Dills, A. H., Alsohim, A. S., McGuffin,
L. J., Studholme, D. J., Silby, M. W., Brockhurst, M. A.,
Johnson, L. J. and Jackson, R. W. (2015) Evolutionary
resurrection of flagellar motility via rewiring of the nitrogen
regulation system. Science, 347 (6225). pp. 1014-1017. ISSN
0036-8075 doi: 10.1126/science.1259145 Available at
http://centaur.reading.ac.uk/39447/
It is advisable to refer to the publisher’s version if you intend to cite from the
work.
Published version at: http://dx.doi.org/10.1126/science.1259145
To link to this article DOI: http://dx.doi.org/10.1126/science.1259145
Publisher: AAAS
All outputs in CentAUR are protected by Intellectual Property Rights law,
including copyright law. Copyright and IPR is retained by the creators or other
copyright holders. Terms and conditions for use of this material are defined in
the End User Agreement .
www.reading.ac.uk/centaur
CentAUR
Central Archive at the University of Reading
Reading’s research outputs online
Title: Evolutionary resurrection of flagellar motility via rewiring of the
nitrogen regulation system
Authors: Tiffany B. Taylor1†, Geraldine Mulley1†, Alexander H. Dills2, Abdullah S.
Alsohim1,3, Liam J. McGuffin1, David J. Studholme4, Mark W. Silby2, Michael A.
Brockhurst5, Louise J. Johnson1,*, Robert W. Jackson1,6
Affiliations:
1
School of Biological Sciences, University of Reading, Whiteknights, Reading, RG6 6AJ,
UK
2
Department of Biology, University of Massachusetts Dartmouth, 285 Old Westport
Road, North Dartmouth, MA 02747, USA
3
Department of Plant Production and Protection, Qassim University, Qassim, Saudi
Arabia P.O. Box 6622
4
College of Life and Environmental Sciences, University of Exeter, Stocker Road,
Exeter, EX4 4QD, UK
5
Department of Biology, University of York, Wentworth Way, York, YO10 5DD, UK
6
The University of Akureyri, Borgir vid Nordurslod, IS-600 Akureyri, Iceland
*Corresponding author: [email protected]
† These authors contributed equally to this work
2
1
One Sentence Summary:
2
Rapid repeatable rewiring of regulatory networks: a nitrogen regulatory gene evolves a
3
new function, restoring flagella to immotile bacteria.
4
5
Abstract:
6
A central process in evolution is the recruitment of genes to regulatory networks. We
7
engineered immotile strains of the bacterium Pseudomonas fluorescens that lack
8
flagella due to deletion of the regulatory gene fleQ. Under strong selection for motility,
9
these bacteria consistently regained flagella within 96 hours via a two-step evolutionary
10
pathway. Step 1 mutations increase intracellular levels of phosphorylated NtrC, a distant
11
homologue of FleQ, which begins to commandeer control of the fleQ regulon at the cost
12
of disrupting nitrogen uptake and assimilation. Step 2 is a switch-of-function mutation
13
that redirects NtrC away from nitrogen uptake and towards its novel function as a
14
flagellar regulator. Our results demonstrate that natural selection can rapidly rewire
15
regulatory networks in very few, repeatable mutational steps.
16
17
18
Main Text:
A longstanding evolutionary question concerns how the duplication and
19
recruitment of genes to regulatory networks facilitates their expansion (1), and how
20
networks gain mutational robustness and evolvability (2). Bacteria respond to diverse
21
environments using a vast range of specialised regulatory pathways, predominantly
22
two-component systems (3), which are the result of adaptive radiations within gene
3
23
families. Due to past cycles of gene duplication, divergence and horizontal genetic
24
transfer, there is often extensive homology between the components of different
25
pathways (4), raising the possibility of cross-talk or redundancy between pathways (5).
26
Here we monitor the recovery of microbial populations from a catastrophic gene
27
deletion: bacteria engineered to lack a particular function are exposed to environments
28
that impose strong selection to re-evolve it, sometimes by recruitment of new genes to
29
regulatory networks (6, 7, 8, 9).
30
In the plant-associated soil bacterium P. fluorescens, the master regulator of
31
flagellar synthesis is FleQ (also called AdnA), a 154-dependent enhancer binding protein
32
(EBP) that activates transcription of genes required for flagellum biosynthesis (10, 11).
33
The starting P. fluorescens strain is AR2; this strain lacks flagella, due to deletion of
34
fleQ, and is unable to move by spreading motility due to mutation of viscosin synthase
35
(viscB), resulting in a distinctive, point-like colony morphology on spreading motility
36
medium (SMM) (12) (Figure 1A). We grew replicate populations of AR2 on SMM; when
37
local nutrients became depleted, starvation imposed strong selection to re-evolve
38
motility. To demonstrate that this finding was not strain-specific, these experiments were
39
replicated in a different strain of P. fluorescens, Pf0-2x. This strain is a 2fleQ variant of
40
Pf0-1, already viscosin-deficient, and is thus unable to move by spreading or swimming
41
motility.
42
After 96 hours incubation of AR2 and Pf0-2x at room temperature on SMM, two
43
breakout mutations were visible conferring first slow (AR2S and Pf0-2xS) and then fast
44
(AR2F and Pf0-2xF) spreading over the agar surface (Fig. 1A). The AR2F strain
45
produces flagella, but we could not detect flagella in EM samples for AR2S (Fig. 1B).
4
46
Genome resequencing revealed a single nucleotide point mutation in ntrB in strain
47
AR2S, causing an amino acid substitution within the PAS domain of the histidine kinase
48
(HK) sensor NtrB (T97P). The fast-spreading strain AR2F had acquired an additional
49
point mutation in the 154-dependent EBP gene ntrC, which alters an amino acid
50
(R442C) within the DNA-binding domain (Table 1 & S2).
51
NtrB and NtrC comprise a two-component system: under nitrogen limitation NtrB
52
phosphorylates NtrC, which activates transcription of genes required for nitrogen uptake
53
and metabolism. To determine how mutations in this separate regulatory pathway
54
restored motility in the absence of FleQ, we performed microarray and qRT-PCR
55
analyses of the ancestral and evolved strains (Fig. S1 & Table S1). The expression of
56
genes required for flagellum biosynthesis and chemotaxis was abolished in AR2
57
compared to wild-type SBW25 (Fig. 2A). The ntrB mutation in AR2S partially restores
58
the expression of flagellar genes, and over-activates the expression of genes involved
59
in nitrogen regulation, uptake and metabolism. The subsequent ntrC mutation in AR2F
60
reduces the expression of nitrogen uptake and metabolism genes, while further up-
61
regulating flagellar and chemotaxis gene expression to wild-type levels (Fig. 2B). While
62
AR2S and AR2F showed higher growth rates than the ancestor in LB medium (the
63
medium on which the mutants arose; Tukey-Kramer HSD test, growth in LB compared
64
to AR2: AR2S, P < 0.001; AR2F, P < 0.001) (Fig. 1C), both mutants grew poorly in
65
minimal medium with ammonium as the sole nitrogen source (Tukey-Kramer HSD test,
66
growth in M9 + ammonium compared to AR2: AR2S, P < 0.001; AR2F, P = 0.001). This
67
is likely to be the result of ammonium toxicity due to the strong up-regulation of genes
5
68
involved in ammonium uptake and assimilation, indicating a pleiotropic cost of this
69
adaptation.
70
Sequencing of the ntrBC locus from independently evolved replicate strains
71
showed that evolution often followed parallel trajectories in both AR2 and Pf0-2x:
72
mutation of ntrB gave a slow-spreading strain and this was followed by mutation of ntrC
73
yielding a fast-spreading strain (Table 1; Fig. 3A & C). While all Pf0-2xF replicates
74
carried mutations in ntrC, several Pf0-2xS strains were not mutated in ntrB, suggesting
75
an alternative evolutionary pathway. Genome resequencing of these strains revealed
76
mutations in glnA or glnK (Table 1; Fig. 3B) likely to result in loss of function leading to
77
abnormally high levels of phosphorylated NtrC: reduced ammonium assimilation by
78
glutamine synthetase (glnA) would impose severe nitrogen limitation in the cell
79
irrespective of nitrogen availability, whereas GlnK is a PII-protein that regulates both
80
NtrB and glutamine synthetase activities.
81
These data suggest a predictable two-step evolutionary process: Step 1
82
increases levels of phosphorylated NtrC, via either (a) a direct regulatory route with
83
mutations in NtrB or GlnK, or (b) a physiological route with loss-of-function mutations
84
reducing glutamine synthetase activity and causing NtrB activation, partially or
85
intermittently reactivating the flagellar cascade. In Step 2, NtrC adapts to enhance
86
activation of the flagellar genes and in doing so, becomes a less potent activator of
87
nitrogen uptake genes. This model explains the microarray data and is consistent with
88
the predicted structural effects of the mutations (Figs. 2 and 3). Specifically, for NtrB the
89
structural changes are likely to either increase kinase or reduce phosphatase activity. In
6
90
support of this, the mutated NtrB(D228A) repeatedly emerging in Pf0-2xS resembles
91
NtrB(D227A) in P. aeruginosa which constitutively activates the Ntr system (13).
92
NtrC is a distant homologue of FleQ, sharing 30% amino acid identity and the
93
same three structural domains (TM-score > 0.7; P < 0.001; Fig. 3D) (14): an N-terminal
94
receiver domain, a conserved central 154-interacting domain, and a C-terminal DNA-
95
binding domain containing a helix-turn-helix (HTH) motif flanked by highly disordered
96
regions. We posit that an overabundance of phosphorylated NtrC activates transcription
97
of flagellar genes through functional promiscuity (15). Consistent with this, the ntrC
98
mutations in fast-spreading strains are predominantly located within or adjacent to the
99
HTH domain and likely influence enhancer-binding specificity; one is a frameshift
100
abolishing the HTH altogether (Fig. 3C). The evolved NtrC’ must be constitutively
101
phosphorylated by overactive NtrB to enable its new multipurpose role, with the result
102
that flagellum biosynthesis and nitrogen regulation are probably no longer responsive to
103
environmental stimuli. Consequently, there is a trade-off between nitrogen utilization
104
and motility (Fig. 1A & C).
105
The flagellar regulatory network may have an unusually dynamic evolutionary
106
history. Flagella are expensive to make, and not always advantageous. Pathogens
107
expressing flagella can trigger strong immune responses in the host, so rapid transitions
108
are seen over short timescales between uniflagellate, multiflagellate and aflagellate
109
states (16). This volatility is reflected in the structure of regulatory networks: in close
110
relatives of Pseudomonas, fleQ appears not to be involved in flagellar gene expression
111
(17), and in Helicobacter pylori, a gene of unknown function can be used as a “spare
112
part” to permit motility in flhB mutants (18). Our results illustrate that trans-acting
7
113
mutations can contribute to gene network evolution (19), but that as predicted, such
114
mutations bear severe pleiotropic costs (20, 21). Genes can retain the potential to take
115
on the functions of long-diverged homologues, suggesting that some degree of
116
evolutionary resilience is a consequence of regulatory pathways that evolve via gene
117
duplication. While de novo origination of new functions in nature is likely to take longer
118
and involve more mutational steps, this system enables us to understand the adaptive
119
process in detail at the genetic and phenotypic level. Here we identified a novel,
120
tractable model for gene network evolution and observed, in real time, the rewiring of
121
gene networks to enable the incorporation of a modified component (NtrC’) creating a
122
novel regulatory function, by a highly repeatable two-step evolutionary pathway with the
123
same point mutations often recurring in independent lineages.
124
125
References:
126
1.
evolution. Annu. Rev. Cell. Dev. Bi. 18, 53-80 (2002).
127
128
2.
3.
133
J. A. Hoch, Two-component and phosphorelay signal transduction. Curr. Opin.
Microbiol. 3, 165-170 (2000).
131
132
M. Aldana, E. Balleza, S. Kauffman, O. Resendiz, Robustness and evolvability in
genetic regulatory networks. J. Theor. Biol. 245, 433-448 (2007).
129
130
J. R. True, S. B. Carroll, Gene co-option in physiological and morphological
4.
S. A. Teichmann, M. M. Babu, Gene regulatory network growth by duplication.
Nat. Genet. 36, 492-496 (2004).
8
134
5.
systems. Annu. Rev. Microbiol. 66, 325-347 (2012).
135
136
E. J. Capra, M. T. Laub, Evolution of two-component signal transduction
6.
Z. D. Blount, C. Z. Borland, R. E. Lenski, Historical contingency and the evolution
137
of a key innovation in an experimental population of Escherichia coli. Proc. Natl.
138
Acad. Sci. USA 105, 7899-7906 (2008).
139
7.
innovation in phage lambda. Science 335, 428-432 (2012).
140
141
8.
B. Hall, The EBG system of E. coli: Origin and evolution of a novel 3galactosidase for the metabolism of lactose. Genetica 118, 143-156 (2003).
142
143
J. R. Meyer, et al., Repeatability and contingency in the evolution of a key
9.
D. Blank, L. Wolf, M. Ackermann, O. K. Silander, The predictability of molecular
144
evolution during functional innovation. Proc. Natl. Acad. Sci. USA 111, 3044-
145
3049 (2014).
146
10.
E. A. Robleto, I. López-Hernández, M. W. Silby, S. B. Levy, Genetic analysis of
147
the AdnA regulon in Pseudomonas fluorescens: Nonessential role of flagella in
148
adhesion to sand and biofilm formation. J. Bacteriol. 185, 453-460 (2003).
149
11.
N. Dasgupta, et al., A four-tiered transcriptional regulatory circuit controls
150
flagellar biogenesis in Pseudomonas aeruginosa. Mol. Microbiol. 50, 809-824
151
(2003).
152
12.
A. S. Alsohim, et al., The biosurfactant viscosin produced by Pseudomonas
153
fluorescens SBW25 aids spreading motility and plant growth promotion. Environ.
154
Microbiol. 16, 2267-2281 (2014).
9
155
13.
W. Li, C.-D. Lu, Regulation of carbon and nitrogen utilization by CbrAB and
156
NtrBC two-component systems in Pseudomonas aeruginosa. J. Bacteriol. 189,
157
5413-5420 (2007).
158
14.
activators. J. Bacteriol. 185, 1757-1767 (2003).
159
160
D. J. Studholme, R. Dixon, Domain architectures of 154-dependent transcriptional
15.
R. Wassem, E. M. de Souza, M. G. Yates, F. D. Pedrosam, M. Buck, Two roles
161
for integration host factor at an enhancer-dependent nifA promoter, Mol.
162
Microbiol. 35, 756-764 (2000).
163
16.
E. Amiel, R. R. Lovewell, G. A. O'Toole, D. A. Hogan, B. Berwin, Pseudomonas
164
aeruginosa evasion of phagocytosis is mediated by loss of swimming motility and
165
is independent of flagellum expression. Infect. Immun. 78, 2937-2945 (2010).
166
17.
R. León, G. Espín, flhDC, but not fleQ, regulates flagella biogenesis in
167
Azotobacter vinelandii, and is under AlgU and CydR negative control.
168
Microbiology 154, 1719-1728 (2008).
169
18.
M. E. Wand, et al., Helicobacter pylori FlhB function: The FlhB c-terminal
170
homologue HP1575 acts as a “spare part” to permit flagellar export when the
171
HP0770 FlhBCC domain is deleted. J. Bacteriol. 188, 7531-7541 (2006).
172
19.
adaptation. Evolution 61, 995-1016 (2007).
173
174
175
H. E. Hoekstra, J. A. Coyne, The locus of evolution: evo devo and the genetics of
20.
G. A. Wray, The evolutionary significance of cis-regulatory mutations. Nat. Rev.
Genet. 8, 206-216 (2007).
10
176
21.
theory of morphological evolution. Cell 134, 25-36 (2008).
177
178
22.
23.
24.
K. Rutherford, et al., Artemis: sequence visualization and annotation.
Bioinformatics 16, 944-945 (2000).
183
184
F. Van Immerseel, et al., Salmonella Gallinarum field isolates from laying hens
are related to the vaccine strain SG9R. Vaccine 31, 4940-4945 (2013).
181
182
M. W. Silby, et al., Genomic and genetic analyses of diversity and plant
interactions of Pseudomonas fluorescens. Genome Biol. 10, R51 (2009).
179
180
S. B. Carroll, Evo-devo and an expanding evolutionary synthesis: a genetic
25.
D. B. Roche, M. T. Buenavista, S. J. Tetchner, L. J. McGuffin, The IntFOLD
185
server: an integrated web resource for protein fold recognition, 3D model quality
186
assessment, intrinsic disorder prediction, domain prediction and ligand binding
187
site prediction. Nucleic Acids Res. 39, W171-W176 (2011).
188
26.
L. J. McGuffin, M. T. Buenavista, D. B. Roche, The ModFOLD4 server for the
189
quality assessment of 3D protein models. Nucleic Acids Res. 41, W368-W372
190
(2013).
191
27.
fold recognition models. Bioinformatics 24, 1798-1804 (2008).
192
193
L. J. McGuffin, Intrinsic disorder prediction from the analysis of multiple protein
28.
D. B. Roche, M. T. Buenavista, L. J. McGuffin, The FunFOLD2 server for the
194
prediction of protein–ligand interactions. Nucleic Acids Res. 41, W303-W307
195
(2013).
196
197
29.
Y. Zhang, J. Skolnick, TM-align: a protein structure alignment algorithm based on
the TM-score. Nucleic Acids Res. 33, 2302-2309 (2005).
11
Acknowledgments:
TBT, LJJ, RWJ and MAB conceived and designed the study. TBT, GM and AA
performed experiments. MWS and AHD performed independent lines of enquiry on Pf02x. DS conducted bioinformatics analysis of genome resequencing data, identified
mutated genes and handled sequencing data. LM conducted the protein structure
prediction and analysis. This work was supported by a Leverhulme grant to LJJ, MAB
and RWJ, BBSRC grant BB/J015350/1 to RWJ, start-up funding from the University of
York to MAB, Qassim University to AA, and Agriculture and Food Research Initiative
Competitive Grant 2010-65110-20392 from the USDA’s National Institute of Food and
Agriculture, Microbial Functional Genomics Program to MWS. TBT, GM, LJJ, RWJ,
MWS, DJS and MAB wrote the paper. We thank Graham Bell, Mark Pagel, Angus
Buckling and James Moir for useful discussions; Peter Ashton for processing of
microarray data; and Konrad Paszkiewicz and Exeter Sequencing Service facility and
support from the following: Wellcome Trust Institutional Strategic Support Fund
(WT097835MF), Wellcome Trust Multi User Equipment Award (WT101650MA) and
BBSRC LOLA award (BB/K003240/1). Sequence data for genomic resequencing of
AR2S and AR2F have been submitted to the SRA under accession numbers
SRR1510202 and SRR1510203, respectively. The eArray design ID for the microarray
is 045642. Microarray data have been submitted to the ArrayExpress database under
accession number E-MTAB-2788 (www.ebi.ac.uk/arrayexpress).
12
Figures 1 – 3:
Fig. 1. Phenotypic assays of motility variants.
A Surface spreading motility assays of ancestral (AR2) and evolved slow spreading
(AR2S) and fast spreading (AR2F) mutants, after 27 hours. B Electron microscopy
confirms the presence of a flagellum in AR2F, but fails to confirm presence in AR2S. C
Mean (N=4) cell doublings per hour (+/- 1 SEM). Strains were grown in differing nitrogen
environments: 10 mM glutamine (Gln), glutamate (Glu) and ammonium (NH4) as the
sole nitrogen source, or in high nutrient lysogeny broth (LB): AR2, F3, 12 = 13.460, P <
0.001; AR2S, F3, 12 = 72.674, P < 0.001; AR2F, F3, 12 = 52.538, P < 0.001. There were
also differences between doubling rate of strains within each nitrogen medium
(Glutamate (Glu), F2, 9 = 12.654, P = 0.002; Ammonium (NH4), F2, 9 = 40.529, P <
0.001), with the exception of Glutamine (Gln) (F2, 9 = 3.703, P = 0.067).
Fig. 2. Heat map of microarray expression profiles for all evolved and ancestral
motility variants.
Heat maps show where there is significant (P 1 0.05) fold-change of 4 2 in genes
selected based on GO-terms for A Bacterial-type flagellum (24 genes) and B Nitrogen
compound transport (146 genes) for all strains. The gradation of colors reflects
normalized raw signal values across the entire array. Genes are ordered according to
chromosomal position to enable clearer visualization of coregulated gene clusters. Full
transcriptome analysis is reported in Supplementary File “Microarray dataset.xlsx”.
Fig. 3. Full chain multi-template 3D models of protein structures of slow and fast
spreading motility variants.
Slow spreading variants can either follow the direct regulatory route through mutation of
NtrB or GlnK (A), or the physiological route through mutation of GlnA causing overactivation of NtrB (B); both routes are predicted to lead to hyperphosphorylation of NtrC.
Fast spreading variants all show mutational changes to NtrC (C), which has a similar
global structure to FleQ (D). The colour scheme represents the variation in models,
13
which correlates with local (per-residue) quality and disorder. Regions coloured in blue
and green represent low local variability in structure, while those in red show high local
variability (see Table 1 and Table S2 for mutation details). § = All mutations mapped
onto SBW25 wildtype protein structures for illustrative purposes; * = Truncated domain.
AR2
14
Slow Spreaders (AR2/Pf0-2xS)
Fast Spreaders (AR2/Pf0-2xF)
Hyperphosphorylation of NtrC
Switch of NtrC-P specificity
ntrB T97P*
ntrC R442C*
ntrB V185K
ntrC K342E
ntrB D179N
ntrC G452R
ntrB L184Q / V185I
ntrC K342V / Frameshift: V342
ntrC N454S
ntrB D228A§
ntrC R441S
ntrC N454S
ntrC P424L
Pf0-2x
ntrC N454S
ntrC L418R
ntrB D228A§
ntrC G414D
ntrC N454S
ntrC F426V
glnK Frameshift: I86*
ntrC R442H*
glnA T237P*
ntrC A445V*
glnA Frameshift: T205*
ntrC R442C*
* = Identified by genome resequencing
§ = Independent ntrB mutant strains, parent to multiple ntrC mutant strains
Table 1. Mutational trajectory towards slow and fast spreading phenotypes.
Mutations confirmed in slow spreading motility variants are predicted to result in
hyperphosphorylation of NtrC; mutations in fast spreading variants lead to predicted
switched specificity of NtrC-P towards FleQ targets. Slow and fast spreading variants
share the same ancestry.
15
Figure 1
16
Figure 2
17
Figure 3
18
Supplementary Materials:
Title: Evolutionary resurrection of flagellar motility via rewiring of the nitrogen regulation
system
Authors: Tiffany B. Taylor1†, Geraldine Mulley1†, Alexander H. Dills2, Abdullah S.
Alsohim1,3, Liam J. McGuffin1, David J. Studholme4, Mark W. Silby2, Michael A.
Brockhurst5, Louise J. Johnson1,*, Robert W. Jackson1,6
Materials and Methods
Fig. S1.
Table S1.
Table S2.
198
Supplementary Materials:
199
200
Materials and Methods:
201
202
Microbiological methods
203
P. fluorescens strains used in this study: SBW25, SBW252fleQ, AR2 (SBW252fleQ IS-
204
5Km-hah: PFLU2552), and Pf0-2x (Pf0-12fleQ). Motility assays are as described in
205
(12). All starting populations were from a single AR2 colony grown on LB agar (1.5%),
19
206
and stab inoculated into the center of a SMM plate using a sterile wire. The initial
207
observed motility mutants (AR2S and AR2F) were cultured immediately, cryopreserved
208
and used for subsequent genome resequencing and microarray analysis. Independently
209
evolved motility mutants were isolated from AR2 and Pf0-2x and cryopreserved.
210
Mutation rates of all lineages were checked by plating an overnight culture onto LB agar
211
supplemented with 100 µg ml-1 rifampicin. All strains were found to have a similar
212
mutation frequency: SBW25 = 6.98 x 10-9 cfu ml-1; 2fleQSBW25 = 3.72 x 10-8 cfu ml-1;
213
AR2 = 2.50 x 10-8 cfu ml-1; AR2S = 2.3 x 10-8 cfu ml-1; AR2F = 1.4 x 10-8 cfu ml-1.
214
M9 nitrogen modified medium (lacking ammonium) was supplemented with 10 mM
215
glutamate, glutamine or ammonium solutions. Strains were grown for 16 hours in LB at
216
27oC and diluted to OD 0.001 in M9 nitrogen modified media, and in LB. Optical density
217
595 nm was measured every 20 minutes, under continuous shaking and at an
218
incubation of 27oC, for 24 hours (Tecan, GENios).
219
220
Molecular methods
221
The mutations present in the original evolved mutants AR2S and AR2F, plus 3 Pf0-2xS
222
and 3 descended Pf0-2xF strains were identified by genome resequencing.
223
Genomic DNA was isolated using Puregene DNA extraction (Qiagen) or Wizard®
224
Genomic DNA Purification (Promega) kits. We resequenced the complete genomes of
225
mutants AR2S and AR2F (original) using the Illumina HiSeq and identified single-
226
nucleotide variants (SNV) by alignment to the SBW25 reference as described previously
227
(22, 23). Further mutants were analyzed by targeted PCR amplification and sequencing
20
228
of ntrBC to determine whether the same two-step process had occurred; primers were
229
designed to flank the region where SNVs had been identified by whole-genome
230
resequencing (NtrC, F: CTTCATCCCCAACTCCTTGA, R:
231
AAGCTGCTGAAAAGCGAGAC; NtrB, F: CTTGCGCCTTGAGTACATGA, R:
232
ATGCGGTCTACCAGGTTACG).
233
234
AR2S and AR2F Isolate Genome Resequencing Details
235
Whole-genome resequencing was performed using the Illumina GA2x. We generated
236
12,065,035 pairs of 36-bp reads for AR2S (SRA accession SRR1510202) and
237
24,075,130 pairs of 36-bp reads for AR2F (SRR1510203). To identify mutations, we
238
aligned the genomic sequence reads against the Pseudomonas fluorescens SBW25
239
genome sequence (NCBI RefSeq accession NC_012660) using BWA-mem version
240
0.7.5a-r405 (http://www.ncbi.nlm.nih.gov/pubmed/19451168). We discarded reads that
241
did not uniquely map to a single location on the SBW25 genome in order to avoid
242
artifacts due to misalignment of sequence reads arising from repeat sequences. The
243
resulting average depths of coverage over the SBW25 genome were 124x and 144x for
244
AR2S and AR2F, respectively. For 99.86% of the SBW25 reference genome sequence
245
(6,713,197 out of 6,722,539 bp) we were able to unambiguously determine the
246
nucleotide sequence in both mutant genomes (on the basis that at these genomic
247
positions there was at least 95% consensus among all aligned reads at those positions
248
for each of the two resequenced genomes and depth of at least 5x). Over the 6,713,197
249
bp of the genome for which we could unambiguously determine the DNA sequence for
250
both mutant genomes, we found only three SNVs with respect to the reference genome
21
251
sequence. Two of these three variants were present in both resequenced genomes: at
252
position 376,439 both resequenced strains had a G whereas in the SBW25 reference
253
genome the base is T. This corresponds to a non-silent change from codon acc to
254
codon Ccc in gene PFLU0344 (glnL/ntrB). The second variant was at position 1,786,536
255
where SBW25 has A but the two resequenced mutants have G; this variant falls in an
256
intergenic region and we have no reason to suspect it has an effect on the phenotypes
257
observed. The third SNV is private to AR2F and falls within the gene PFLU0343
258
(glnG/ntrC). In SBW25 and AR2S the base is G but in AR2F it is A; this results in a non-
259
silent change from codon cgt to codon Tgt. Aside from these three variant sites, the
260
remainder of the 6,713,197 bp of unambiguously resolved genome sequence contained
261
no variation from the SBW25 reference genome.
262
263
Pf0-2xS and Pf0-2xF Isolate Genome Resequencing Details
264
Library preparation was performed using Illumina Nextera XT kit with Nextera XT
265
Indexes and sequenced as 250PE reads from MiSeq.
266
Whole-genome resequencing of Pf0-2x strains was performed using the Illumina MiSeq.
267
We generated 1,145,122 forward and 1,141,017 reverse reads of 251-bp for Pf0-2xS_1,
268
1,279,360 forward and 1,272,990 reverse reads of 251-bp for Pf0-2xF_1.1, 1,962,339
269
forward and 1,941,676 reverse reads of 251-bp for Pf0-2xS_2, 907,328 forward and
270
906,083 reverse reads of 251-bp for Pf0-2xF_2.1, 935,862 forward and 932,415 reverse
271
reads of 251-bp for Pf0-2xS_3, and 1,194,665 forward and 1,190,975 reverse reads of
272
251-bp for Pf0-2xF_3.1. These genomic sequence reads were mapped against the
273
Pseudomonas fluorescens Pf0-1 genome sequence (NCBI RefSeq accession
22
274
NC_007492.2) using CLC Genomics Workbench 6.0; unmapped reads were not
275
included in further analysis. The average coverage depths for each library were: 44.6 for
276
Pf0-2xS_1, 49.8 for Pf0-2xF_1.1, 76.1 for Pf0-2xS_2, 35.3 for Pf0-2xF_2.1, 36.4 for Pf0-
277
2xS_3, and 46.5 for Pf0-2xF_3.1. Out of the 6,438,405 bp Pf0-1 genome, 97.06-99.35%
278
of nucleotides (depending on sample) had a minimum 95% consensus among all
279
mapped reads. Using probabilistic variant detection to investigate unambiguous
280
nucleotide positions in each sequenced strain, we identified two SNVs and a 3bp
281
deletion common to all mutants (Table S2). These three common changes were also
282
present in the parental strain (Pf0-2x) and were thus not considered further. A 15bp
283
deletion was identified in both the Pf0-2xS1 and F1.1 strains with a unique SNV
284
occurring in the latter. An additional SNV was common to Pf0-2xS2 and Pf0-2xF2.1, yet
285
another to both Pf0-2xS3 and Pf0-2xF3.1, and a single unique SNV was discovered in
286
Pf0-2xF2.1 and another in Pf0-2xF3.1 with respect to the reference genome sequence.
287
288
The presence of glnA, glnK and ntrC mutations in resequenced strains was confirmed
289
by PCR of candidate regions from parental and evolved strains followed by Sanger
290
sequencing of the amplicons. Mutations in ntrB were detected by targeted PCR
291
amplification and sequencing, and ntrC mutations in derivatives of ntrB mutants were
292
detected the same way. In some cases, multiple primers were used to cover the full
293
length of the gene. This enabled us to be certain about the mutations, and rule out any
294
others. Primers used for Pf0-1 were: ntrB, F: ATTGCGCCTCGAGTACATGA and
295
TCGACACGGTTTCACTACGG, R: ATGCGGTCGACCAGATTGCG and
296
TCGGAGCCTTGGTTTGGTTT; ntrC, F: AAGCTGCTGAAAAGCGAGAC and
23
297
GATTAAGGGTCACGGTGCCT, R: CTTCATGCCGAGTTCCTTGA and
298
CACTGGAACAAGGAGCCACA; glnA, F: GGAGGCCTTTCTTTGTCACG and
299
ACGCTTGTAGGAGTTGGTGG and CGGGAAGTAGCCACCTTTCA, R:
300
CCACCAACTCCTACAAGCGT and CAAGTCCGACATCTCCGGTT and
301
CTACCCGCCCTAATTCACCC; glnK, F: GCCGGGCATTACGATAGACA, R:
302
TGCGCTTAGACTTGAGTCGG.
303
304
Microarray design
305
For microarray analysis, RNA was extracted from SBW25, SBW252fleQ, AR2, AR2S
306
and AR2F using an RNeasy kit (Qiagen) and assayed for quality [2100 Bioanalyzer
307
(Agilent), and Nanodrop 1000 Spectrophotometer (Thermo Fisher Scientific)]. RNA was
308
labelled, hybridised to custom Agilent arrays and scanned according to the
309
manufacturer’s instructions (Technology Facility, Dept. of Biology, University of York,
310
UK). Differentially expressed genes were identified by ANOVA using the Benjamini-
311
Hochberg FDR correction. No commercially available microarrays were available for P.
312
fluorescens, so a custom design was created using the Agilent eArray system
313
(http://earray.chem.agilent.com/earray). The P. fluorescens SBW25 genome was
314
loaded into Artemis (24) (http://www.sanger.ac.uk/resources/software/artemis), and all
315
open reading frames marked as CDSs and these putative CDS written to a FASTA file
316
which was uploaded to eArray. A probeset was created containing five unique probes
317
per CDS, plus the appropriate Agilent control sequences. The probes were then laid out
318
onto a standard Agilent 8x60K format slide. Microarrays were validated using qRT-PCR
319
(MyiQTM, Bio-Rad).
24
320
321
RNA labelling and hybridisation
322
RNA isolated from each sample was labelled with Cy-3 using the Agilent Low Input
323
Quick Amp WT Labelling Kit (one color) according to the manufacturer’s instructions.
324
Briefly, the RNA is reverse transcribed using a primer mix containing oligo-dT and
325
random nucleotide primers containing T7 promoters, then the cRNA amplified using the
326
T7 polymerase in the presence of Cy-3 labelled dCTP. The labelled cDNA was then
327
quality controlled and only samples with a specific activity of > 6 were used hybridised
328
to the arrays.
329
Samples were hybridised to the arrays, incubated and washed in accordance with the
330
manufacturer’s instructions, and the slides scanned using an Agilent Microarray
331
Scanner and the fluorescence quantified with the Feature Extraction software. The
332
resulting files were then loaded into GeneSpring for further analysis.
333
334
Initial Data Analysis
335
GeneSpring software version 12.6 was used for all data analysis. Initial analysis was
336
performed to identify those genes that were differentially expressed between any
337
conditions in the experiment, followed by Gene Set Expression Analysis (GSEA) on the
338
basis of Gene Ontology terms. Briefly, the sample replicates were grouped, and then
339
ANOVA performed to identify differentially expressed genes, using the Benjamini-
340
Hochberg FDR p-value correction (with a FDR of 5%), and a Tukey HSD post-hoc test
341
used for each gene to identify which samples were significantly different from each
25
342
other for that gene. Those GO categories that were over-represented in the set of
343
differentially expressed genes were then identified.
344
345
3D protein models
346
The IntFOLD server (25) version 2 was used to build multi-template 3D models from the
347
wild type and mutant sequences of AdnA, NtrB, NtrC, FleQ, GlnA and GlnK. The 3D
348
models were quality checked using the ModFOLD4 protocol (26). Intrinsic protein
349
disorder was predicted using the DISOclust method (27). Binding sites were predicted
350
using the FunFOLD2 method (28). Models were structurally aligned and scored using
351
TM-align (29).
352
353
Further Discussion of Microarray Results
354
Full transcriptome analysis is reported separately in Supplementary File “Microarray
355
dataset.xlsx”. When genes were filtered for significant fold changes (P 1 0.05) of 4 2,
356
and sorted by descending fold changes, the genes that showed the greatest changes in
357
expression are those highlighted in this study. Specifically, when comparing AR2 and
358
AR2S, the large majority of genes with greatest up regulation are related to nitrogen
359
transport, whereas when comparing AR2S to AR2F, the large majority of genes are
360
related to flagellar motility and chemotaxis.
361
Our microarray data show that the average level of flagella gene expression in AR2S is
362
between 1.5-2-fold lower than in SBW25, but considerably higher than AR2 (between 3
363
and 75 fold). There are at least two possible models to explain why we were unable to
26
364
observe flagella in AR2S: (i) Phosphorylated NtrC (NtrC-P) only weakly interacts with
365
flagella gene promoters, but the over-abundance of NtrC-P in AR2S saturates NtrC-
366
dependent promoters and the excess NtrC-P drives expression of flagella genes. The
367
level of transcription is suboptimal, thus the flagella may be unstable and support limited
368
or transient motility; (ii) Promiscuous activity of over-abundant NtrC-P (e.g. interacting
369
with RpoN from solution) leads to expression from flagella promoters, but this is not
370
equivalent in all cells and only a proportion of the population express sufficient levels to
371
produce a functional flagellum at any one time.
27
372
Fig. S1: Validation of gene expression changes detected by microarray using
qRT-PCR.
A – E Box plots show gene expression, relative to wild-type SBW25, in genes related to
nitrogen utilization and flagellar function. Note, data presented on log scales.
28
A. SBW251fleQ
Bacterial-type
Flagellum
Regulation of
Nitrogen
Utilization
GENE
SYMBOL
flgF
Pflu4448
(Flagellin)
glnA
ntrB
ntrC
GENE CHIP ARRAY
Fold Change
p-value
-79.06
1.52E-07
REAL TIME PCR
Fold Change
p-value
-8.4
0.196
-158.33
2.09E-07
-22.59
0.002
1.01
-1.54
-1.74
2.24E-06
1.37E-07
1.58E-07
9.14
-4.5
-1.39
0.187
0.41
0.136
B. AR2
GENE
SYMBOL
Bacterial-type
Flagellum
Regulation of
Nitrogen
Utilization
flgF
Pflu4448
(Flagellin)
glnA
ntrB
ntrC
GENE CHIP ARRAY
Fold
p-value
Change
-76.48
1.52E-07
REAL TIME PCR
Fold Change
p-value
-2.84
0.265
-128.22
2.09E-07
-6.61
0.004
1.03
-1
-1.41
2.24E-06
1.37E-07
1.58E-07
3.27
-2.24
-1.31
0.393
0.424
0.231
C. AR2S
GENE
SYMBOL
Bacterial-type
Flagellum
Regulation of
Nitrogen
Utilization
flgF
Pflu4448
(Flagellin)
glnA
ntrB
ntrC
GENE CHIP ARRAY
Fold
p-value
Change
-2.53
1.52E-07
REAL TIME PCR
Fold Change
p-value
3.13
0.192
-1.71
2.09E-07
4.34
0.042
6.08
28.18
27.14
2.24E-06
1.37E-07
1.58E-07
25.49
26.45
49.14
0.008
0.017
0.059
D. AR2F
Bacterial-type
Flagellum
Regulation of
Nitrogen
Utilization
GENE
SYMBOL
flgF
Pflu4448
(Flagellin)
glnA
ntrB
ntrC
GENE CHIP ARRAY
Fold Change
p-value
1.1
1.52E-07
REAL TIME PCR
Fold Change
p-value
4.49
0.1
3.6
2.09E-07
16.57
0.065
11.34
58.27
63.72
2.24E-06
1.37E-07
1.58E-07
175.63
96.09
49.73
0.005
0.006
0.03
Table S1: Mean fold change (relative to SBW25) from qRT-PCR and microarrays
with p-values (A – D).
Colours correspond to genes with significant ( 4 2 fold change) up regulated (red) and
down regulated (blue) changes.
29
Strain
AR2S_1
AR2F_1.1
AR2S_2
AR2F_2.1
AR2S_3
AR2F_3.1
AR2S_4
AR2F_4.1
Nucleotide change
T376439G*
G374322A*
G376175A/
T376174A
T374622C
G376193A
C374291G
G376176A
T376174A
T374622C/
I A374625
66431771-3*
AA change
T97P
R442C
Gene
ntrB
ntrC
Domain/Function
PAS domain
HTH domain
V185K
ntrB
HK domain
K342E
D179N
G452R
L184Q
V185I
K342V
FS from V342
2G85
ntrC
ntrB
ntrC
HTH domain
HK domain
Miscellaneous function
ntrB
HK domain
ntrC
HTH domain
gidB
66171472-87*
C382879T*
66431771-3*
FS from I86
R442H
2G85
glnK
ntrC
gidB
T388861G*
6902225*
G382870A*
66431771-3*
T237P
FS from S544
A445V
2G85
glnA
carB
ntrC
gidB
6388958*
G382880A*
A384603C
A382843G
C382883A
A382843G
C382933T
A382843G
A384603C
T382951G
G382963A
A382843G
T382928G
FS from T205
R442C
D228A
N454S
R441S
N454S
P424L
N454S
D228A
L418R
G414D
N454S
F426V
glnA
ntrC
ntrB
ntrC
ntrC
ntrC
ntrC
ntrC
ntrB
ntrC
ntrC
ntrC
ntrC
Methyltransferase
domain
PII regulator
HTH domain
Methyltransferase
domain
Catalytic domain
Miscellaneous function
HTH domain
Methyltransferase
domain
Catalytic domain
HTH domain
Miscellaneous function
HTH domain
HTH domain
HTH domain
HTH domain
HTH domain
Miscellaneous function
Miscellaneous function
Miscellaneous function
HTH domain
HTH domain
Pf0-2xS_1
Pf0-2xF_1.1
Pf0-2xS_2
Pf0-2xF_2.1
Pf0-2xS_3
Pf0-2xF_3.1
Pf0-2xS_4
Pf0-2xF_4.1
Pf0-2xF_4.2
Pf0-2xF_4.3
Pf0-2xF_4.4
Pf0-2xF_4.5
Pf0-2xS_5
Pf0-2xF_5.1
Pf0-2xF_5.2
Pf0-2xF_5.3
Pf0-2xF_5.4
* = Identified by genome resequencing.
Table S2: Full details of mutations identified in strains with parental AR2 and Pf02x origin.
Numbers before decimal points represent independent lineages, and numbers after
represent derived strains. Note fast spreading mutants contain all nucleotide changes
within a lineage.